Google has released Gemini 3.8 Flash, a general reasoning and coding model that arrives only three weeks after 3.7 Flash. Google reports a 73.7% score on DeepSWE v1.1 for long-running software-engineering tasks, up from 65.3% for its predecessor and close to the 74% result it lists for Claude Opus 5.
Introductory pricing remains $0.75 per million input tokens and $3.75 per million output tokens, with standard rates scheduled to double in January 2027. The unchanged token price does not mean unchanged task cost. Google says 3.8 Flash takes additional reasoning steps and calls tools iteratively on difficult work, and independent evaluator Artificial Analysis estimates $0.58 per task versus $0.40 for 3.7 Flash.
Developers can access the model through Google AI Studio, Antigravity and Android Studio, while businesses get it through Gemini Enterprise. It is also reaching the Gemini app, AI Mode in Search and Google Sheets for paying subscribers. Google recommends lower reasoning settings or the older model where efficiency matters most, making workload-level testing more useful than comparing headline token rates alone.